According to the resources, missions and restraints of the data relay satellite optical network, a scheduling algorithm based on improved self-adaptive genetic algorithm was put forwarded. Considering the multi-relay satellite, multi-window, multi-optical-antenna and multi-mission PRI, the model was established. The missions were scheduled by the scheduling operates: the ascertainment of current mission scheduling time and the refreshment of latter mission time-window. The whole weight of the scheduled missions was set as the cost value and the scheduling schemes were optimized by the self-adaptive genetic algorithm. The simulation scene including 3 relay satellites, 12 user satellites, 6 antennas and 60 missions, the result reveals that the algorithm obtains satisfactory results in both time and optimization which is suitable in multi-user, multi-mission and multi-optical-antenna recourse scheduling.